JOURNAL OF APPLIED PHYSICS VOLUME 85, NUMBER 1 1 JANUARY 1999 Pure nuclear Bragg reflection of a periodic 56Fe/57Fe multilayer L. Dea´ka) KFKI Research Institute for Particle and Nuclear Physics, P.O.B. 49, H-1525 Budapest, Hungary G. Bayreuther Institut fu¨r Angewandte Physik, Universita¨t Regensburg, Universita¨tsstr. 31, D-93053 Regensburg, Germany L. Bottya´n KFKI Research Institute for Particle and Nuclear Physics, P.O.B. 49, H-1525 Budapest, Hungary E. Gerdau II. Institut fu¨r Experimentalphysik, Universita¨t Hamburg, Luruper Chaussee 149, D-22761 Hamburg, Germany J. Korecki Department of Solid State Physics, University of Mining & Metallurgy, al. Mickiewicza 30, PL-30-059 Krako´w, Poland E. I. Kornilov Frank Laboratory of Neutron Physics, Joint Institute for Nuclear Research, 141 980 Dubna, Moscow Region, Russia H. J. Lauter Institute Laue-Langevin, B.P. 156, F-38042 Grenoble Cedex 9, France O. Leupold II. Institut fu¨r Experimentalphysik, Universita¨t Hamburg, Luruper Chaussee 149, D-22761 Hamburg, Germany D. L. Nagy KFKI Research Institute for Particle and Nuclear Physics, P.O.B. 49, H-1525 Budapest, Hungary A. V. Petrenko and V. V. Pasyuk-Lauterb) Frank Laboratory of Neutron Physics, Joint Institute for Nuclear Research, 141 980 Dubna, Moscow Region, Russia H. Reuther and E. Richter Forschungszentrum Rossendorf e.V., Institut fu¨r Ionenstrahlphysik und Materialforschung, Pf. 510119, D-01314 Dresden, Germany R. Ro¨hlobergerc) II. Institut fu¨r Experimentalphysik, Universita¨t Hamburg, Luruper Chaussee 149, D-22761 Hamburg, Germany E. Szila´gyi KFKI Research Institute for Particle and Nuclear Physics, P.O.B. 49, H-1525 Budapest, Hungary Received 3 March 1998; accepted for publication 3 October 1998 Grazing incidence nuclear multilayer diffraction of synchrotron radiation from a periodic stack of alternating 56Fe and 57Fe layers was observed. Resonant layer fraction, substrate size, flatness, and surface roughness limits were optimized by previous simulations. The isotopic multilayer ML sample of float glass/57Fe 2.25 nm / 56Fe 2.25 nm /57Fe 2.25 nm 15/Al 9.0 nm) nominal composition was prepared by molecular beam epitaxy at room temperature. Purity structure and lateral homogenity of the isotopic ML film was characterized by magnetometry, Auger electron, Rutherford backscattering, and conversion electron Mo¨ssbauer spectroscopies. The isotopic ML structure was investigated by neutron and synchrotron Mo¨ssbauer reflectometry. Surface roughness of about 1 nm of the flat substrate curvature radius 57 m was measured by scanning tunneling microscopy and profilometry. A pure nuclear Bragg peak appeared in synchrotron Mo¨ssbauer reflectometry at the angle expected from neutron reflectometry while no electronic Bragg peak was found at the same position by x-ray reflectometry. The measured width of the Bragg peak is in accordance with theoretical expectations. © 1999 American Institute of Physics. S0021-8979 99 09201-4 a Electronic mail: deak@rmki.kfki.hu b Present address: Universita¨t Konstanz, Postfach 5560, D-78434 Konstanz, Germany. c Present address: Universita¨t Rostock, August-Bebel-Str. 55, D-18055 Rostock, Germany. 0021-8979/99/85(1)/1/7/$15.00 1 © 1999 American Institute of Physics 2 J. Appl. Phys., Vol. 85, No. 1, 1 January 1999 Dea´k et al. I. INTRODUCTION spectrum. The angular acceptance of a single crystal nuclear Bragg monochromator is 20 rad. Nonresonant nested In nuclear resonant scattering NRS of synchrotron ra- channel-cut monochromators have the suitable angular ac- diation SR low-lying levels of an ensemble of identical ceptance their energy bandwidth being, however, 1 meV. nuclei are coherently excited by the synchrotron radiation Two kinds of nuclear monochromators of suitable bandwidth pulse. Since the levels are, as a rule, split by hyperfine inter- and angular acceptance have been suggested so far, viz. actions, the spatial and temporal coherence of the scattering grazing incidence antireflection GIAR films6,7 and syn- results in characteristic patterns both of the angular distribu- thetic periodic multilayers. Synthetic periodic multilayers tion and the time evolution of the scattered radiation, which should have a nuclear Bragg reflection at a glancing angle bear simultaneous and correlated information about topology which is well above the critical angle of the electronic total and internal fields in the sample under study. SR is scattered reflection so that the electronic reflectivity of the mirror at both by nuclei and by electrons and these two processes in- is low enough. Both kinds of narrow band monochromators terfere with each other, as well. have roughly the same bandpass and reflectivity irrespective Conventional Mo¨ssbauer spectroscopy and NRS of SR of the particular multilayer design, so that the real differ- are, although delivering similar information on hyperfine in- ences in existing multilayer devices are of technical nature. teraction and lattice dynamics, complementary rather than The fabrication both of GIAR films and of synthetic pe- equivalent to each other. The principal difference is that riodic multilayers is difficult due to various technological when the energy spectrum is scanned by the Doppler-shifted problems. The substrate should be extremely flat and the radiation of a source, the recorded signal presents the in- roughness of the surface as well as of the interfaces should coherent sum of the spectral components of the transmitted be suppressed, the same holding true for a possible interdif- radiation. In case of time domain NRS of SR, however, the fusion of the layers. Also a sufficient lateral homogeneity of response is formed by the coherent sum of the spectral com- the mirrors should be ensured. In what follows we shall re- ponents of the scattered radiation. From practical points of strict ourselves to the 14.413 keV resonance of 57Fe. view however, NRS of SR has various advantages. Due to the high brilliance, the low divergence, the high degree of Pure nuclear reflections are expected from two kinds of polarization and the pulsed time structure of SR, Mo¨ssbauer- synthetic periodic multilayers. In the first group the hyperfine like measurements on extremely small samples have become interaction in different layers has a superstructure as com- possible. Further, the performance of grazing incidence NRS pared to the interlayer spacing. The second group comprises experiments can be improved by several orders of magnitude synthetic periodic multilayers built from alternating resonant by using SR rather than radioactive sources, a fact opening and nonresonant isotopes of the same element synthetic iso- the way to a new kind of depth selective hyperfine spectros- topically periodic multilayers . Typical examples of the first copy, viz. synchrotron Mo¨ssbauer reflectometry SMR .1­4 class are antiferromagnetically coupled multilayers. Pure NRS experiments at synchrotron beamlines are now rou- nuclear reflections of SR from an Fe/Cr multilayer with a Cr tinely performed in a forward scattering, b grazing inci- thickness of 1.0 nm mediating an antiferromagnetic interac- dence, and c Bragg reflection geometries. The high count tion between the Fe layers have been observed by Toellner et rate of the direct beam and of the prompt electronic scatter- al.8 Recently, pure nuclear reflections of antiferromagnetic ing poses major detector dead time problems in forward scat- origin have been observed on Fe/FeSi multilayers.9 tering experiments. SMR is usually performed close to or The idea of building nuclear monochromators from syn- even below the critical angle of the electronic total reflection thetic isotopically periodic multilayers was first suggested by where, due to the interplay of electronic and nuclear scatter- Trammell et al.10 and also discussed later by Kabannik.11 In ing, a considerable amount of delayed photons along with a a short communication,12 Kikuta et al. reported on the fact of high intensity of electronically reflected prompt photons are having observed the pure nuclear reflection of the available.3,5 Nuclear Bragg reflections may also electroni- 14.413 keV radiation of a 57Co source from a cally be allowed, a fact leading to the same detection prob- 56Fe 1.2 nm /57Fe 0.8 nm)] n synthetic isotopically peri- lem. One of the keys to high performance NRS experiments odic multilayer. Unfortunately, very few details of sample with SR is the development of monochromators with an en- preparation and measurement are given. The systematic de- ergy bandpass in the order of the hyperfine interaction en- sign and development of synthetic isotopically periodic ergy i.e., 1 eV . A narrower bandwidth leads to a partial multilayer monochromators started in 1991 when Chumakov excitation of the hyperfine levels of the probe nuclei and a and Smirnov13 optimized the total thickness t and bilayer response time comparable to the nuclear lifetime while a thickness d as well as the ratio of the thickness of the broader one unnecessarily overloads the detector with non- resonant layer to d. Since then, there have been various at- resonant radiation. tempts vide infra to prepare synthetic isotopically periodic The angular acceptance of the monochromator should multilayers with the sputtering technique and to observe their not be less than the divergence of the synchrotron radiation pure nuclear reflections. beam typically of 20­60 rad at bending magnets and un- The fabrication of 56Fe/57Fe multilayers with sputtering dulators and several hundreds of rad at wigglers. The band- had failed14 due to interdiffusion of 56Fe and 57Fe into each width of 1 eV can be achieved by use of a pure nuclear other. Conversion electron Mo¨ssbauer spectra CEMS of Bragg reflection of a single crystal due to the hyperfine in- such samples showed -iron, with some asymmetric broad- teraction of the nuclei delivering a rather complicated energy ening of the lines towards lower magnetic fields.14 J. Appl. Phys., Vol. 85, No. 1, 1 January 1999 Dea´k et al. 3 Chumakov et al.15 reported the observation of the different from that of Chumakov and Smirnov.13 While these nuclear reflection of 14.413 keV radiation from a 57Co authors optimized the reflectivity R(E, ) r(E, ) 2 as av- source, using a synthetic periodic multilayer eraged around the resonant energy E0 over an energy inter- 57Fe 2.0 nm /Sc 3.3 nm)] 20 on an extremely smooth val of 1 eV at the Bragg angle B and neglected the glass substrate. Due to the lack of isotopic periodicity of electronic scattering we suggested that, for time domain ex- iron, the reflection was also electronically allowed. The iron periments, it was more reasonable to optimize the integral. layers were amorphous and nonmagnetic. Later, a pure t nuclear reflection of the 57Co gamma radiation was B /2 2 D B Rt t, dt d , observed16 using a synthetic isotopically periodic multi- B /2 t1 layer 57Fe 2.2 nm /Sc 1.1 nm /Fe 2.2 nm /Sc 1.1 nm)] with R 25. Clearly, the electronic Bragg peak of this mirror lies at t(t, ) rt(t, ) 2 where rt is the time domain reflec- tivity amplitude, the Fourier transform of the energy domain double the angle of the pure nuclear reflection. The first at- reflectivity amplitude r(E, ), is the divergence of the tempt to see the pure nuclear reflection of synchrotron radia- synchrotron radiation, while t tion from a synthetic isotopically periodic multilayer was 1 and t2 are the limits of the time window set to detect the delayed photons. With t made by Chumakov et al.,17 using this 57Fe/Sc/Fe/Sc mir- 1 12 ns, t ror. In this experiment the nuclear Bragg peak was seen, and 2 175 ns, and 50 rad and accounting for nuclear and electronic scattering D( the applicability of the mirror as a nuclear monochromator B) was found to be maximum for 0.4¯0.5 depending on the number of bi- was demonstrated. Nevertheless, as seen in Fig. 1 of the layers, i.e., at values somewhat higher than 0.33 as cal- corresponding paper,17 a weak electronic Bragg peak of re- culated by Chumakov and Smirnov.13 flectivity at the 10 3 level also appeared together with the The proper selection of the material and of the dimen- nuclear one. The first demonstration of a pure nuclear sions of the substrate is a crucial problem. A surface and reflection from a synthetic isotopically periodic multilayer interface roughness of R having no electronic Bragg peak at all was by Ro¨hlsberger q 1.5 nm will considerably attenu- ate or even completely smear out both the structural Bragg et al.,18 using a 57Co source and a multilayer peak and the Kiessig oscillations related to the total thickness 56Fe5B4C 4.8 nm /57Fe5B4C 3.9 nm)] 10. of the multilayer. Clearly, the surface roughness of the sub- In an attempt to test the capabilities of molecular beam strate should by no means exceed this value. Commercial Si epitaxy MBE in fabricating eV monochromators for wafers and cleaved single crystals often have R nuclear resonant scattering experiments at synchrotrons, we q 1.5 nm. Therefore we decided to use a high quality float glass sub- prepared a synthetic isotopically periodic multilayer from al- strate with a nominal R ternating layers of 56Fe and 57Fe. After extensive character- q 0.5 nm. The optimum length l of the substrate is given by l z/ ization of the sample with various methods we found a pure B where z is the height of the synchrotron beam 1 mm at beamline F4 of HASY- nuclear reflection of SR. LAB in order not to lose intensity due to grazing incidence geometry. Unfortunately, as we shall see, the resulting l II. EXPERIMENT 10 cm cannot easily be realized and a considerable loss in A. Sample design intensity should be accepted. The flatness i.e., the long scale curvature radius c) of When designing the nuclear resonant mirror we focussed the sample is given by the condition that the angle of the on three items: a to optimize the structure of the multilayer substrate plane normal between two opposite ends of the so that the reflectivity of the monochromator at the nuclear sample should be less than or approximately equal to an Bragg peak should be maximum b to select a flat substrate acceptable fraction say 20% of the expected width of the of low surface roughness, and c to choose a preparation Bragg peak B , in other words c l/ B . Taking l method by which interdiffusion, enhanced interface rough- 10 cm, B 1 mrad, and 0.2, we have c 500 m, a ness, and amorphization can be avoided and a sufficient lat- condition not easy to fulfill. For a substrate of a given cur- eral homogeneity can be ensured. vature c the condition for the length l c B can be The bilayer thickness d of the multilayer was selected used. following the criteria of Chumakov and Smirnov13 viz. by Another and opposite limit for l is given by the prepara- prescribing a low (10 2¯10 3) electronic reflectivity at the tion conditions in view of the demand of a sufficient lateral expected Bragg angle. The nominal bilayer thickness d was homogeneity of the layer thickness vide infra and by other chosen to be 4.5 nm corresponding to a Bragg angle B technical limitations of a given evaporation chamber. Be- ( /2d)2 2c 10.3 mrad with 0.0860 nm of the sides, the substrate should be thick enough in order to pre- 14.413 keV radiation and c 3.8 mrad the critical angle of vent the sample from bending. With all these in mind, the electronic total reflection of -iron for this wavelength . At dimensions of the float glass substrate were selected to be 15 10.3 mrad the electronic reflectivity of a semi-infinite mm 15 mm 1 mm. With l 15 mm, the condition for the -iron has been calculated13 with the Fresnel formula to be curvature c 75 m was set. 10 3. In order to avoid interdiffusion of the two iron isotopes The ratio of the thickness of the resonant layer to d during preparation we decided to use a preparation technique was optimized utilizing the technique of characteristic by which the energy of the deposited atoms can be well matrices;19 more details of the numerical technique were controlled and also a proper crystallinity of the layers is ex- given elsewhere.2,3 Our optimization criterium was slightly pected. In view of the difficulties in previous attempts to 4 J. Appl. Phys., Vol. 85, No. 1, 1 January 1999 Dea´k et al. prepare synthetic isotopically periodic multilayers by sput- tering or by conventional vacuum evaporation techniques, we decided to use the MBE technique. One of the basic problems of MBE is to optimize the distance h of the sub- strate from the evaporation cells. On one hand, if h is too high, the deposition yield of the expensive isotope 57Fe will be low; on the other hand, if h is too low, the lateral homo- geneity of the layer thickness will suffer. We note that, as a rule, it is quite difficult to ensure a controlled 57Fe deposition yield with magnetron sputtering. Clearly, the relative lateral layer thickness inhomogeneity should be less than or about FIG. 1. Conversion electron Mo¨ssbauer spectrum of the 56Fe/ 57Fe B / B 2% with the above values otherwise the Bragg peak will be smeared out. Simple geometrical consid- multilayer at room temperature taken in a flow of 96 vol. % He and 4 vol. % CH eration sets the condition l/h 2 4 counter gas with a 57Co RH source. Velocity scale is given relative to B / B, provided that -Fe at room temperature. B / B 1. With the above values we have h 5.4 cm. Ar ion beam at a current density of about 0.5 mA/mm2. The B. Sample preparation and characterization Auger peaks Fe(LM M) at 703 eV, O(KLL) at 507 eV, and The synthetic isotopically periodic multilayer sample C(KLL) at 266 eV were used for the analysis. The observed of nominal structure float glass/57Fe(2.25 nm)/ profiles see Fig. 2 indicated no oxidization in the majority 56Fe(2.25 nm)/57Fe(2.25 nm) 15/Al(9.0 nm) sample of the iron layers. Carbon appeared only in the uppermost 1 ``56Fe/57Fe multilayer,'' for later use was grown with MBE nm of the multilayer and the uppermost 5 nm turned out to in a ultrahigh vacuum UHV deposition chamber built at the be oxidized. Another 5¯10 nm of iron at the Fe/glass inter- University of Mining & Metallurgy, Krako´w, Poland. The face seemed to be oxidized, as well. No significant Auger distance between the evaporation cells and the substrate was peak of Al could be identified. We explain this fact by the h 7 cm. The base pressure during evaporation was better already mentioned inhomogeneity a strong wedge-like than 10 8 Pa, the deposition rate was 0.8 nm/s. The BeO shape of the Al coating thickness. Indeed, in order not to crucibles of the Knudsen cell contained 57Fe of 95% enrich- damage the mirror, Auger profiling was done close to one ment and 56Fe. During deposition, the float glass substrate corner of the sample. was held at room temperature. The substrate was heated prior The short scale surface roughness of the sample was to preparation to about 150 °C. The evaporation rate was measured by scanning tunnel microscopy on a Rasterscope monitored with a quartz gauge with an accuracy correspond- 3000 instrument. Long scale surface roughness data were ing to 0.01 nm. The average thickness of a single isotope Fe also extracted from surface profilometry. Both methods re- layer was kept within an accuracy better than 0.1 nm as sulted in roughness values of Rq 1 nm. compared with the nominal one. The Fe layer structure was The total amount of iron at different beam spots was coated with Al in order to prevent the iron film from oxidiz- determined by Rutherford backscattering spectrometry ing. As we shall see, the Al coating turned out to be inho- RBS using a 1.62 MeV 4He beam obtained from a 5 MeV mogeneous. This fact, however, had no serious influence on Van de Graaff accelerator of the KFKI Research Institute for the reflectivity of the sample. Particle and Nuclear Physics, Budapest. An ion current of The flatness of the sample was determined by a Dektak typically 20 nA as measured by a transmission Faraday cup20 8000 surface profilometer. The long scale curvature radius was used. Figure 3 shows the measured RBS spectrum taken c turned out to be not less than 57 m, roughly fulfilling the at the spot closest to the center of the sample ``central spec- requirement c 75 m. Both in situ and conventional CEMS showed the pres- ence of pure -iron. The resonance lines, as shown in Fig. 1, had no significant broadening, their relative intensities being 2.80:3.42:1.00, i.e., close to 3:4:1 characteristic for the mag- netization lying in the sample plane. The hyperfine field Hhf 33.0 1 T corresponds to that of -Fe within the ex- perimental error. Vibrating sample magnetometry also sup- ported the presence of -iron with a coercivity field of 30 Oe at room temperature. Sputter depth profiling was performed with a scanning Auger electron spectrometer MICROLAB 310F. The rela- tively low value of c and d prevented us from exactly mea- suring the depth of the sputtering crater with the profilome- ter. Nevertheless, the obtained depth about 60 nm was still FIG. 2. Sputtering Auger electron spectroscopy depth profile of the in fair accordance with the nominal value of 72.8 nm. Sput- 56Fe/ 57Fe multilayer. The calibration of the depth scale was determined by tering was performed on the rotating sample with a 3 keV the total iron thickness of 82.6 nm as taken from neutron reflectometry data. J. Appl. Phys., Vol. 85, No. 1, 1 January 1999 Dea´k et al. 5 FIG. 4. Measured spin dependent neutron reflectivities R and R of the 56Fe/ 57Fe multilayer as a function of the momentum transfer Q. The fit curves correspond to the float glass/57Fe 4.6 nm / 56Fe 2.9 nm / 57Fe 2.3 nm 15/Al 5.7 nm structure. FIG. 3. RBS spectrum taken close to the center x 8.5 mm, y 7.0 mm of the 56Fe/ 57Fe multilayer using a 50 mm2 ORTEC surface barrier detector at 165° scattering angle with a solid angle of 5.5 msr. The sample was tilted at 60°. The arrows show the energy belonging to the respective elements on the atomic volume and, (z) is the average magnetic mo- the surface. The inset displays the Fe and Al thicknesses at ten different ment per atom at depth z]. For ferromagnetic films magne- spots normal and italic numbers, repsectively . tized in the sample plane no spin-flip processes take place and the functional F can easily be calculated using the tech- trum'' as well as the simulated one determined by the RBX nique of characteristic matrices.19,22 program21 using a layer structure of float glass/Fe 81 nm / The neutron reflectometry measurements of the 56Fe/57Fe Al 3 nm . The peak areas of Fe and Al as a function of beam multilayer were made at the time-of-flight polarized neutron position were used to create a ``homogeneity map'' of the spectrometer SPN at the pulsed reactor IBR-2 at the Joint sample inset in Fig. 3 . Normalization was performed to the Institute for Nuclear Research at Dubna.23 The neutron respective peak areas of the central spectrum. Unfortunately, wavelength ranged from 0.08 to 0.12 nm. Experimental spin- the expectation for the lateral homogeneity of the iron thick- dependent reflectivities R and R were obtained by apply- ness to be 2% has not been fulfilled, its overall value being ing the saturating external magnetic field of 0.1 T parallel to 15%. Nevertheless, about one half of the sample (x the film surface and are shown together with the fit to the 8 mm; see inset turned out to be good enough for the data in Fig. 4. The oscillations correspond to a total thickness synchrotron measurements. of the sample of 88.3 nm. A Bragg peak appears around the momentum transfer value Q 1.2 nm 1 both on the R and R reflectivity curves. The Bragg peak position is deter- C. Polarized neutron reflectometry mined by the thickness of the isotope bilayer structure. To The only established method capable of testing the qual- establish the scattering length density in the float glass a ity of synthetic isotopically periodic multilayers is neutron reflectivity curve of substrate was measured separately. The reflectometry. In fact, in contrast to x rays, the nuclear scat- degree of polarization of the neutrons has been indepen- tering length of neutrons is different for different isotopes of dently measured and has properly been accounted for. From the same chemical element. Therefore, a structural Bragg the fit to the data we determined the structure and the inter- peak at about the same values of the scattering vector mo- facial roughness of the film. The layer structure was fitted to mentum transfer Q should appear from synthetic isotopi- be float glass/57Fe 4.6 nm / 56Fe 2.9 nm /57Fe 2.3 nm cally periodic multilayers in neutron reflectometry and in 15/Al 5.7 nm) along with the roughness values Rq of 2.0, SMR experiments while no such peak is expected in non- 1.0, and 1.8 nm on the surface, at the interfaces and on the resonant x-ray reflectometry due to the homogeneous elec- substrate, respectively. The thickness higher than nominal of tronic density of the sample. Furthermore, neutron reflecto- the 57Fe layer near substrate is probably due to the presence metry with polarized neutrons is capable of revealing the of oxygen and to a small diffusion of Fe into the glass. magnetic structure of the film. The objective of the polarized neutron reflectometry ex- periment is to obtain the reflectivities R and R for up and D. Synchrotron Mo¨ssbauer reflectometry down spin neutrons with respect to the applied magnetic SMR experiments were performed at the nuclear reso- field as a function of Q. The measured reflectivities R (Q) nance scattering beamline F4 of HASYLAB, DESY, Ham- are functionals of the nuclear and magnetic neutron scatter- burg. The 14.413 keV resonant radiation was selected with a ing length profiles bn(z) and bm(z), respectively, along the Si 111 high heat load monochromator and a Si 4 2 2 / depth z, so that R (Q) F (bn(z)/V bm(z)/V) with Si 12 2 2 nested channel-cut monochromator. The height bm(z) C (z), where Q 4 sin / and C r0 /2 of the synchrotron radiation beam was adjusted to 60 m, 0.269 542 10 12 cm/ B where is the neutron wave- the vertical divergence was about 20 rad. The beam width length, r0 is the classical electron radius, 1.91304 is the was adjusted between 1 and 4 mm according to the experi- magnetic moment of the neutron in nuclear magnetons, V is mental requirements vide infra . 6 J. Appl. Phys., Vol. 85, No. 1, 1 January 1999 Dea´k et al. film. The Bragg peak is due to the isotopic periodicity of the multilayer. Its presence shows that no significant interdiffu- sion took place in the synthetic isotopically periodic multilayer prepared by using the MBE technique. The fact that the R and R Bragg peaks in Fig. 4 appear almost exactly with the same intensity, shows that the SR was purely nuclear diffracted by the isotope periodicity of the multilayer. The slight deviation in the Q values of the Bragg peak measured with neutrons and photons is due to the dif- ference in the critical angles of the total reflection as well as to the fact that with neutrons the whole area while with SR only a part of the somewhat inhomogeneous sample was measured. The continuous curve on Fig. 5 b corresponds to D( ) t2tRt(t, )dt. Rt(t, ) is calculated using an optical 1 model4 and with the provision that there was no preferred inplane direction of the magnetization within the film. For the calculation the layer parameters determined from neutron reflectivity fit was used. A universal roughness parameter was found from the fit to be Rq 1.2 nm in accordance with the neutron data. The width of the Bragg peak also fairly accords with the expectations. FIG. 5. Nonresonant x-ray curve a and time integral delayed resonant reflectivity b of SR from the 56Fe/ 57Fe multilayer. Delayed counts were collected in the time interval 17 ns t 146 ns at glancing angle corre- sponding to momentum transfer Q. The continuous curves were calculated from the layer parameters of the neutron reflectivity fit for the float glass/ 57 IV. CONCLUSION Fe 4.6 nm / 56Fe 2.9 nm /57Fe 2.3 nm 15/Al 5.7 nm structure. We reported on the observation of a pure nuclear reflec- In view of the lateral homogeneity data, as inferred from tion of synchrotron radiation from a synthetic isotopically the RBS measurements, the most homogeneous part of the periodic multilayer prepared by MBE. This technique has sample had to be selected. First the multilayer was oriented proved to be suitable for fabricating synthetic isotopically with the y axis Fig. 3 parallel to the beam. This direction periodic multilayers with no sign of layer interdiffusion. This enables one to illuminate the largest possible homogeneous is probably due to the fact that the target temperature is bet- part of the sample. Afterwards, the most homogeneous part ter controlled in MBE than in sputtering or conventional of the sample (x 8 mm) was selected by subsequent x-ray vacuum evaporation chambers. Using MBE for the prepara- reflectometry measurements performed at parallel strips of tion of synthetic isotopically periodic multilayers the crystal- the sample with a beam of 1 mm width. It was established line and magnetic structure of the phase of iron is main- that the region is homogeneous enough to observe the Bragg tained. The Bragg peak was strongly attenuated due to the peak. Then nuclear resonant scattering experiment was done roughness effects. only on this part of the sample with a beam width of 4.0 mm. The main purpose of the study described here was to For technical reasons, it was not possible to apply an external demonstrate the capabilities of MBE in fabricating narrow magnetic field. A series of time integral rocking curves was band monochromators rather than to build an operating de- recorded and added (t1 17 ns, t2 131 ns). The sum of the vice. For the latter purpose isotopically periodic but elec- measurements taken in a run of about 10 h is shown in Fig. tronically homogeneous paramagnetic multilayers of broad 5. The first peak at Q 0.56 nm 1 is due to the interplay of resonance lines are, admittedly, better suited. A pure ferro- the electronic and nuclear scattering at the critical angle of magnetic -iron multilayer has, however, the advantages that the electronic total reflection.3,5 The second one is the Bragg the experiment can be well compared with theory, the hyper- peak that was found on the resonant reflectivity curve almost fine interaction in the film being well defined. In contrast to exactly at the same value of the momentum transfer (Q a pure nuclear reflection of antiferromagnetic origin8,9 that 1.25 nm 1) as in the neutron experiment. At the same can be switched in a high external magnetic field the pure time, no Bragg peak can be seen on the nonresonant x-ray nuclear peak resulting from isotopic periodicity is not sensi- reflectivity curve Fig. 5 . Due to the limited intensity avail- tive to external fields. Further, in low external fields the able at the beamline, it has not been possible to record a time alignment of the sublayer magnetizations of an antiferromag- spectrum at the nuclear Bragg peak. netic multilayer can be quite complicated,9 a feature not ex- pected for soft ferromagnetic materials like -iron. III. DISCUSSION In conclusion, the present study is a contribution to de- The difference in spin-dependent neutron reflectivities veloping efficient eV monochromators for nuclear resonant R and R in Fig 4 is due to the ferromagnetic nature of the scattering experiments at synchrotrons. J. Appl. Phys., Vol. 85, No. 1, 1 January 1999 Dea´k et al. 7 ACKNOWLEDGMENTS 8 T. S. Toellner, W. Sturhahn, R. Ro¨hlsberger, E. E. Alp, C. H. Sowers, and This work was partly supported by the PHARE AC- E. E. Fullerton, Phys. Rev. Lett. 74, 3475 1995 . 9 L. Bottya´n, J. Dekoster, L. Dea´k, A. Q. R. Baron, S. Degroote, R. Moons, CORD Program under Contract No. H-9112-0522 and by the D. L. Nagy, and G. Langouche, Hyperfine Interact. 113, 295 1998 . Hungarian Scientific Research Fund OTKA under Contract 10 G. T. Trammell, J. P. Hannon, S. L. Ruby, P. Flinn, R. L. Mo¨ssbauer, and Nos. 1809, T016667, and F022150. The neutron reflectom- F. Parak, AIP Conf. Proc. 38, 46 1977 . etry measurement at JINR was partly supported by the 11 V. A. Kabannik, Version of a Nuclear Resonance Filter for Mo¨ssbauer Dubna Fund of the Hungarian Government and by the IN- Diffraction with Synchrotron Radiation in Russian , preprint Institute of TAS Grant No. 93/3617. The authors also thankful for the Nuclear Physics, 1989 . 12 partial support of the Deutsche Forschungsgemeinschaft, the S. Kikuta, X. Zhang, Y. Yoda, K. Izumi, and T. Ishikawa, Rev. Sci. Instrum. 60, 2126 1989 . Deutscher Akademischer Austauschdienst DAAD , the 13 A. I. Chumakov and G. V. Smirnov, Pis'ma Zh. Eksp. Teor. Fiz. 53, 258 Hungarian Academy of Sciences and the Polish Academy of 1991 JETP Lett. 54, 271 1991 . Sciences in frames of bilateral projects. The Auger electron 14 A. I. Chumakov private communication . spectroscopy, scanning tunelling microscopy and profilom- 15 A. I. Chumakov, G. V. Smirnov, S. S. Andreev, N. N. Salashchenko, and etry measurements were part of Grant No. ERBCI- S. I. Shinkarev, Pis'ma Zh. Eksp. Teor. Fiz. 54, 220 1991 JETP Lett. PACT922051 of the Commission of the European Commu- 55, 509 1992 . 16 nities. Helpful discussions with Dr. A.I. Chumakov are A. I. Chumakov, G. V. Smirnov, S. S. Andreev, N. N. Salashchenko, and S. I. Shinkarev, Pis'ma Zh. Eksp. Teor. Fiz. 55, 495 1992 JETP Lett. gratefully acknowledged. 54, 216 1991 . 17 A. I. Chumakov, G. V. Smirnov, A. Q. R. Baron, J. Arthur, D. E. Brown, 1 D. L. Nagy and V. V. Pasyuk, Hyperfine Interact. 71, 1349 1992 . S. L. Ruby, G. S. Brown, and N. N. Salashchenko, Phys. Rev. Lett. 71, 2 L. Dea´k, L. Bottya´n, and D. L. Nagy, in Condensed Matter Studies by 2489 1993 . Nuclear Methods, Proc. XVIII, Zakopane School of Physics, Zakopane, 18 R. Ro¨hlsberger, E. Witthoff, E. Lu¨ken, and E. Gerdau, J. Appl. Phys. 74, 1993, edited by E. A. Gorlich and K. Tomala Institute of Nuclear Phys- 1933 1993 . ics, Krako´w, 1993 , p. 269. 19 M. Born, E. Wolf, Principles of Optics Pergamon P, Oxford, 1970 , p. 3 L. Dea´k, L. Bottya´n, and D. L. Nagy, Hyperfine Interact. 92, 1083 1994 . 51. 4 L. Dea´k, L. Bottya´n, D. L. Nagy, and H. Spiering, Phys. Rev. B 53, 6158 20 F. Pa´szti, A. Manuaba, C. Hajdu, A. A. Melo, and M. F. da Silva, Nucl. 1996 . 5 A. Q. R. Baron, J. Arthur, S. L. Ruby, A. I. Chumakov, G. V. Smimov, Instrum. Methods Phys. Res. B 47, 187 1990 . 21 and G. S. Brown, Phys. Rev. B 50, 10354 1994 . E. Ko´tai, Nucl. Instrum. Methods Phys. Res. B 85, 588 1994 . 6 22 J. P. Hannon, N. V. Hung, G. T. Trammell, E. Gerdau, M. Mueller, R. J. Penfold and R. K. Thomas, J. Phys.: Condens. Matter 2, 1369 1990 . Ru¨ffer, and H. Winkler, Phys. Rev. B 32, 5068 1984 . 23 D. A. Korneev, V. V. Pasyuk, A. V. Petrenko, and E. B. Dokukin in 7 J. P. Hannon, G. T. Trammell, M. Mueller, E. Gerdau, R. Ru¨ffer, and H. Surface X-ray and Neutron Scattering, edited by H. Zabel and I. K. Rob- Winkler, Phys. Rev. B 32, 6363 1985 . inson Springer, Berlin, 1992 , p. 213.